SUMMARY Among various approaches to restore vision, including optogenetic stimulation and stem cell therapy, only the electrode-based retinal prosthesis has validated its clinical promises. However, it suffers from fundamental limitations: it requires a complicated surgery for device implantation, has both the limited number and fixed location of stimulation sites, and above all, has a low spatial resolution since electric currents spread in conductive media like the retina. A decade ago, photothermal stimulation with infrared light opened the possibility of ?remotely? activating neurons without the aid of optogenetics, but the strong water absorption of infrared light leads to bulk tissue heating and associated adverse effects. To enable cellular-resolution, ?remote? neural activation without the bulk heating, we have demonstrated that a combined use of gold nanoparticles and near- infrared light (negligibly absorbed by water) can produce highly-localized heat via surface plasmon resonance, and this can activate neurons by generating capacitive membrane currents and/or opening temperature-sensitive ion channels. We also have shown that appropriate chemical conjugation of nanoparticles further enhances the efficacy of near-infrared stimulation. This promising neuromodulation approach, however, has yet not demonstrated its potential as a retinal prosthesis. Here, we propose to develop, optimize, and validate this novel technology, termed plasmonic retinal prosthesis, and compose its potential with several important advantages when compared to the electrode-based retinal prostheses: (1) it does not require any device implantation but only involves intravitreal injection of gold nanorods (AuNRs); (2) the single-cell resolution can be achieved in vivo; (3) stimulation locations or targeted ganglion cells are freely adjustable; (4) the number of activatable neurons per unit time can be as high as 100,000 neurons per second (in our pilot setup); and (5) the performance is further upgradable after ?installation? as the relevant technologies advance because every key component locates outside the eye. We will develop this promising technology through theoretical study, ex vivo optimization, in vivo validation, and long-term testing. First, since it is essential in any novel neural interface to have an accurate model of the system in order to optimize the design, we will advance our mathematical neuron model to investigate two mechanisms currently under debate and determine the initial parameters for the following animal experiments (Aim 1). Next, using our custom experimental setup that integrates a scanning laser system and fluorescence microscope, we will develop and optimize single-cell stimulation of retinal ganglion neurons in retina explants of mice with genetically-encoded Ca2+ indicators, followed by both the demonstration of patterned multi-neuron stimulation and the optimization of AuNR chemistry (Aim 2). Finally, we will integrate our experimental and theoretical work to validate in vivo that patterned near-infrared stimulation of the retina induces neural activation in the visual cortex similar to natural visual stimuli, with the parameters being further optimized, and will perform a longitudinal experiment to observe and quantify its long-term efficacy and toxicity (Aim 3).